Tissue and protein hydrolysates have been routinely used as a source of peptides in cell culture media since the late 1800's. They are the most common undefined culture media component in present use in bacteriology and often replace serum in mammalian culture (S. Saha and A. Sen. 1989. Acta Virol. 33:338–343). Hydrolysates and serum are not optimal sources of peptides for culture media, however, because their compositions are undefined and variable, and serum may harbor pathogens such as BSE.
It has been recognized that peptides are generally preferred nutrients as compared to their constituent amino acids. Several approaches have been taken in an effort to determine which specific peptides are utilized by a cell culture as a means for identifying defined peptides which affect growth or some other biological activity. A common practice is to analyze spent media in an attempt to identify compounds which have been consumed during culture. This seldom leads to a single compound which can be isolated and studied. The spent media approach cannot identify compounds which affect the cell and are not removed from the medium (e.g., signaling compounds). In an alternative approach, specific proteins are digested and the HPLC-purified peptide fragments are spiked back into the medium to evaluate their effects. This approach may identify a peptide which performs better than the whole protein digest or tissue digest, but the number of possible peptides for analysis is limited. For example, it has been reported that casein hydrolyzed by the neutral protease of Micrococcus caseolyticus (M. J. Desmazeaud and J. H. Hermier. 1972. Eur. J. Biochem. 28:190–198) and a papain digest of glucagon (M. J. Desmazeaud and J. H. Hermier. 1973. Biochimie 55:679–684) enhance the growth of Streptococcus thermophilus. In this case the stimulatory peptides were isolated and characterized. It has also been found that trypsin digested K-casein enhances the growth of the genus Bifidobacterium (M. Poch and A. Bezkorovainy. 1991. J. Agric. Food Chem. 39:73–77), however, the specific peptides which produce this result were not identified. Azuma, et al. (1984. Agric. Biol. Chem. 48:2159–2162) and Bezkorovainy, et al. (1979. Am. J. Clin. Nutr. 32:1428–1432) reported that a glycopolypeptide derived from enzyme-digested human casein promotes growth of Lactobacillus bifidus. Tryptic fragments of human β-casein have also been reported to stimulate DNA synthesis in BALB/c3T3 cells (N. Azuma, et al. 1989. Agric. Biol. Chem. 53:2631–2634). The sequences of two such tryptic fragments were determined. These prior art methods are time consuming and have resulted in identification of only a few peptides which generate marginal improvements. Their success has primarily been limited by the raw materials, which are restricted by the starting substrates and the digesting agents used in their preparation.
More recently, developments in peptide synthesis technology have made it at least plausible to prepare and screen large numbers of compounds for media enhancement, either as individual defined sequences or as a mixture of variable sequences in a peptide library. Many of these sequences would not be present or detected in the traditional undefined materials. The library approach has provided an opportunity to screen more peptide sequences for desired biological effects in cell culture, but the primary methods have major disadvantages. Assaying compounds individually means screening millions of samples containing randomly-generated sequences. As a practical matter, the exhaustive synthesis and screening of libraries is often prohibitively expensive and also time-consuming. The combinatorial approach runs the risk of missing potential lead compounds due to poor representation (low concentration) of each compound in the cocktail due to solubility restraints and masking effects that can occur from competing compounds. In an effort to reduce the number of sequences which must be screened in a library, practitioners have “fixed” certain residue positions during synthesis of the library. That is, certain residues at certain positions in the sequence are added randomly, but the residues at other positions are defined. Such a synthetic peptide library is described in U.S. Pat. No. 5,556,762. In addition, Geysen (WO 86/00991) describes libraries comprising peptide sequences which are a combination of defined and undefined amino acid residues. Furka, et al. (1988. 14th Int Cong. Biochem. Vol. 5, Abst. FR:013) discloses relatively simple mixtures of tetrapeptides in which the N- and C-terminal residues are fixed and any one of three residues occur at each position in between. Fodor, et al. (1991. Science 251:767–773) teach solid phase peptide synthesis on slides, using predetermined amino acids coupled to defined areas of the slide using photomasks. In this way an array of 1024 different peptides with defined C-termini was synthesized. All of these techniques attempt to circumvent the individual screening of millions of peptides and to increase the amount of a given sequence in the library to simplify screening and identification of biologically active peptides.
While fixed-position (i.e., limited diversity) libraries reduce the number of sequences which must be screened, they also limit the number of different sequences available for screening and thus may reduce the probability of identifying a sequence with the desired properties. In the publications discussed above, there was typically no attempt made to “re-expand” the number of available sequences in order to identify additional sequences which may have properties similar to those in the limited diversity library. Recently, information about the properties of a compound identified in a more limited library of sequences has been used to generate a more diverse library of compounds which are structurally similar to the initial compound identified. These additional compounds, which were not present in the initial library, may exhibit biological activities which are similar to the initial lead compound. This approach is often referred to as rational design of targeted libraries. See, for example, S. Cho, et al. 1998. J. Chem. Inf. Comput. Sci. 38:259–268.
For media applications, simply identifying a compound delivering the desired enhancement is not sufficient. To impact the overall media optimization process, a lead compound must be rapidly scaled up and made available in a time frame which will impact the typical media optimization cycle. Further, the method used for the initial scale up must be in-line with the planned commercial manufacturing process capable of delivering the compound at a cost in-line with benefit. The ideal discovery process would link the initial library design to the preferred manufacturing process and thereby avoid a series of subsequent libraries aimed at finding compounds with similar performance attributes that can also be manufactured. None of the existing fixed-library designs address this need. Indeed, the manufacturing aspect is left to chance.
There is therefore a need in the art for chemically-defined peptides with well-characterized biological activities which can be added to culture media to produce a desired biological effect. Such peptides reduce the number and quantity of undefined components in culture media, reduce the need for animal-derived components, improve media consistency and quality control and provide a means for precisely controlling and adjusting performance of the cell culture. The present invention employs a peptide library approach to select and identify peptides which meet these needs, in particular a process that links discovery and manufacturing of peptides which affect cell growth (either positively or negatively) or which enhance or inhibit cellular protein production.